Currently, the predominant methods to label and/or to conjugate molecules to proteins, especially, when small-molecule payloads or labels are concerned, involve the chemical conjugation with specific linker molecules that covalently attach the payload to free lysine and/or cysteine amino acids of the proteins.
However, many proteins, like e.g. antibodies that are of particular interest for immunotargeting strategies, are fairly large proteins, that may contain several lysine and cysteine residues. Because linker-mediated, chemical conjugation is a stochastic process, linker-mediated chemical ligation of payloads leads to heterogeneous mixtures of conjugated proteins that may differ in their therapeutic efficacy and/or diagnostic utility. Obviously, mixtures of protein-payload conjugates also represent a significant challenge in the regulatory approval process for therapeutic conjugates, as batch-to-batch variation and/or variations in the active pharmaceutical ingredient (API) are negatively viewed by regulatory authorities due to potential safety concerns
In addition, if a defined ratio of payload to protein is desired, it is often necessary to purify the conjugate with the desired conjugation stoichiometry. This is not only tedious, but can significantly add to the cost-of-goods in the manufacturing process, as often only a fraction of the linker-mediated conjugated protein represents the desired ratio of payload conjugation. This is particularly true for therapeutically relevant antibody/drug conjugates (ADCs), where depending on the toxin employed, 3 to 4 toxin molecules appear to be advantageous, but antibodies with no toxin coupled to up to 8 toxins per antibody coupled are found in typical linker-mediated chemical conjugation reactions (Panowski et al. (2014)).
Despite of the limitations described above, all antibody/drug conjugates currently in clinical trials, or approved by the health authorities for the therapy of disease, have been generated by linker-mediated chemical conjugation of toxic small-molecule drugs to antibodies (Lambert (2012) or Mullard (2013)).
It is widely acknowledged in the industry and by scientific experts in the field, that site-specific and stoichiometric conjugation of molecular payloads, including toxin or label molecules to immunoligands would have significant advantages in comparison to chemical, linker-mediated conjugation. This is evidenced by attempts to target the chemical conjugation to specific amino acids in the protein structure (Panowski et al. (2014)).
On one hand, this is attempted by mutating certain positions in the protein structure to delete unwanted and/or to provide desired conjugation sites (i.e. lysine and/or cysteine residues) to which the linker-ligation can be targeted (McDonagh et al. (2006) or Junutula et al. (2008)).
On the other hand, control of chemical conjugation to proteins is attempted by incorporation of unnatural amino acids at certain positions, like selenocysteine, p-azidophenylalanine, or acetylphenylalanine (Hofer et al. (2009), Axup et al. (2012), or Lemke (2011)).
However, all of these approaches change the primary amino acid sequence of the protein to be conjugated, and may result in undesired functional properties. Furthermore, the incorporation of unnatural amino acids, as described above, is often low efficient, and does not allow for a quantitative incorporation of specific labelling sites to proteins.